Documentation

Learn how to use Firecase to spin up instant cloud sandboxes for AI agents and developers.

What is Firecase?

Firecase is a cloud platform for spinning up instant, isolated Linux sandboxes. Built on Firecracker microVMs, each sandbox boots in under 150ms and provides complete hardware-level isolation.

Key Features

Millisecond Boot Times — VMs start in under 150ms, perfect for on-demand code execution.
Hardware Isolation — Each sandbox runs in its own microVM with a dedicated kernel, providing true isolation.
Instant Checkpoints — Snapshot VM state and restore instantly. Fork from any checkpoint to parallelize work.
Full Linux Environment — Run any binary, install any package. SSH access, web terminal, and complete API control.
Flexible Resources — Scale from 1 to 32 vCPUs and up to 64GB RAM per sandbox.

Use Cases

AI Code Execution — Let AI agents write and execute code safely in isolated environments.
Development Environments — Spin up pre-configured dev environments instantly.
Security Testing — Analyze malware or test exploits without risking your infrastructure.
CI/CD Pipelines — Run builds and tests in isolated, reproducible environments.

How It Works

Firecase manages the full lifecycle of cloud sandboxes through a simple API:

Create an instance — A persistent sandbox definition with a name, resource profile, and base image.
Start the VM — Boot a Firecracker microVM in under 150ms. The VM gets its own kernel, filesystem, and network.
Execute code — Run commands, read/write files, open terminals, and forward ports — all via REST API or WebSocket.
Checkpoint & fork — Snapshot the running VM state. Restore later, or fork to create parallel branches.
Stop & resume — Stop the VM to save resources. All disk state persists to S3. Resume anytime.

Bash

# Create a sandbox
curl -X POST https://api.firecase.ai/instances \
  -H "Authorization: Bearer $FIRECASE_API_KEY" \
  -d '{"name": "my-sandbox"}'

# Start the VM (boots in ~150ms)
curl -X POST https://api.firecase.ai/instances/{id}/vm \
  -H "Authorization: Bearer $FIRECASE_API_KEY"

# Execute code
curl -X POST https://api.firecase.ai/instances/{id}/exec \
  -H "Authorization: Bearer $FIRECASE_API_KEY" \
  -d '{"command": ["python3", "-c", "print(\"Hello from Firecase!\")"]}'

Architecture Overview

Every Firecase sandbox is a Firecracker microVM running on bare metal servers with NVMe storage and dedicated CPU cores. The storage layer uses a copy-on-write chunk model backed by S3, enabling instant forking and persistent state across VM restarts.

Component	Role
Control Plane	REST API, scheduling, quota management, manifest operations
Compute Agent	Per-server VM lifecycle, process spawning, gRPC bridge
Storage Daemon	Block device emulation, multi-tier caching, WAL durability
Guest Agent	In-VM command execution, file I/O, terminal multiplexing

Ready to get started?

Follow the Quickstart Guide to create your first sandbox in under 5 minutes.

Quickstart